Cause and Effect – An Agent-Based Approach to Simulate Strategic Level Driver Behavior

Abstract

Contemporary traffic simulation systems are able to reproduce traffic situations in a highly realistic fashion. Most of these systems capture the physics of relevant elements and use these models to reproduce traffic scenes of selected roads, cities, and even entire countries. All too often, these models neglect human factors, yet, is it not that human beings significantly determine the outcome of traffic situations? The aim of this thesis is twofold. The first objective is to determine to what extend current traffic simulations are able to reproduce human behavior in traffic situations. For this purpose, psychological behavior conceptualizations are analyzed and compared to the expressiveness of those models that are currently implemented in computer-aided traffic simulation frameworks. The comparison reveals that there are in fact deficiencies - especially in the representation of one particular form of behavior, namely strategic level driver behavior. This particular form of behavior, however, has a serious impact on traffic situations and is therefore of great importance for traffic simulation systems, respectively. Identified deficiencies mainly concern the representation and implementation of those factors that, following human factors psychology, significantly affect strategic level driver behavior. The second objective of this work is to correct this misalignment and to present a simulation model which reproduces strategic level driver behavior in compliance with psychological findings. This work substantiates the thesis that such a simulation model can be realized by means of agent-technology and the service metaphor.

@PhdThesis{Lutzenberger2015Cause,
  Title                    = {Cause and Effect -- An Agent-Based Approach to Simulate Strategic Level Driver Behavior},
  Author                   = {Marco L"{u}tzenberger},
  School                   = {Technische Universit"{a}t Berlin},
  Year                     = {2015},
  Month                    = {February},
  Type                     = {Doctoral Dissertation},

  Abstract                 = {Contemporary traffic simulation systems are able to reproduce traffic situations in a highly realistic fashion. Most of these systems capture the physics of relevant elements and use these models to reproduce traffic scenes of selected roads, cities, and even entire countries. All too often, these models neglect human factors, yet, is it not that human beings significantly determine the outcome of traffic situations?

The aim of this thesis is twofold. The first objective is to determine to what extend current traffic simulations are able to reproduce human behavior in traffic situations. For this purpose, psychological behavior conceptualizations are analyzed and compared to the expressiveness of those models that are currently implemented in computer-aided traffic simulation frameworks. The comparison reveals that there are in fact deficiencies---especially in the representation of one particular form of behavior, namely strategic level driver behavior. This particular form of behavior, however, has a serious impact on traffic situations and is therefore of great importance for traffic simulation systems, respectively. Identified deficiencies mainly concern the representation and implementation of those factors that, following human factors psychology, significantly affect strategic level driver behavior.

The second objective of this work is to correct this misalignment and to present a simulation model which reproduces strategic level driver behavior in compliance with psychological findings. This work substantiates the thesis that such a simulation model can be realized by means of agent-technology and the service metaphor.},
  Doi                      = {10.14279/depositonce-4392}
}
Author:
Marco Lützenberger
Category:
Dissertation
Year:
2015
Location:
Doctoral Dissertation, Technische Universität Berlin, Berlin, Germany